31 | From Weeks to Minutes: An AI-Powered Step-by-Step Blueprint to Radically Boost Marketing Efficiency, with Ashley Gross, head of Global Campaigns at Commercetools

25:39But then also I said review their LinkedIn profiles. So I am going to copy it in twice. Again, just to be very transparent with what I'm expecting it to do. and then here is what it generated so if you're on this call Mallory Phillips Thomas McDonald Debra Murphy Andrew Anderson and Larissa Marktier these are the random names that I picked to be a part of this so here are the activities again just feel free to let me know if these are inaccurate or accurate I would love to know and I can regenerate them if they're not here we go photography outdoor adventure gourmet cuisine location scouting a coastal retreat with advantages and disadvantages urban metropolis again advantage and disadvantage and then countryside retreat and then it actually gives me its opinion so based on the considerations above so it already went through the advantages and disadvantages and decided what was the best option for me so that I didn't have to.

26:40Again, seconds and takes all the thinking out of it. It picked urban metropolis, the advantages of easy accessibility, abundant accommodations, and a diverse range of attractions. And then, yeah, should you have any further questions or require additional information, please do not reach out. This is a travel manager. They get paid$150 for a vet to do this, and I just did it in seconds for free. This is amazing. I can regenerate it again. If anybody is selected, I can't see the chat, but if anyone's in there saying, those are not happy, so I'm happy to regenerate this. I want to say something as far as the overall process.

27:15And first of all, explain how insane this is. And second, generalize it to other processes that you can do today with similar platforms. So the first thing is, Ashley mentioned several different people that you would usually either employees of the company or people that you would hire that are expert on something specific in order to do this entire process. So from the research of defining the client personas through the steps of then finding the actual people who would fit that persona within, in this case, we've done the whole US, but we could have identified, we just want to do this around New York City or Texas or any other location, Southwest.

27:57So you can define an area or a region or a city where you want to do this and then only look at profiles where you have offices or where you want to grow your new customer share or whatever the case may be. It gives you a list of those people, which would have been a while to create manually because you have to go through multiple lists of people that you either buy or that you curate through LinkedIn on your own And then you have to go and do the tedious work of finding what these people care about. So you can find hobbies that they all share. Why? Because that's your attraction for the event.

28:38So yes, it's a business networking event. But what if all these people like to play pickleball? And you can bring one of the top 10 people in pickleball in the world to take out how is pickleball connected to the way they're successful in business. and now you have an awesome event because anybody who's on the list who cares about pickleball will most likely want to come to the event and now there's the actual event planning of location and what you want to do and i think that's probably the next step which is what's in the event itself right yep so schedule a conference and a common geographical location which i already found and create an event with a budget experience and schedule i'm going to show you its full potential So I'm actually not going to limit it with budget.

29:21But this is really nice when you're on a tight budget like we all are all the time. I'm going to actually give it a bigger budget just for fun, just to see what all it comes up with. And then I'm going to say the metropolis area because that's what it recommended. That's all I'm going to do. So this prompt is Actors, Event, Player, Plan a Professional Development Conference for people in the, I'm going to say technology industry with a budget of$150 ,000 in a metropolis area, which aims to attract high profile and emerging entrepreneurs. I'm saying start researching the most suitable venues with adequate space and amenities.

29:58Create a list of potential keynote speakers, panel members, and breakout session facilitators. Your event should include thought provoking discussions, interactive group activities, and hands-on product demonstrations. You can plug in your product here or you can just leave it in parentheses if you're going to use it in a chat function like this because it will remember it. I also asked it to create a detailed schedule for the event, including exact timings and locations to keep the event running smoothly. This is going to be a hefty output, but this is it. This is the end game. So I like that it starts out with an email too.

30:30Save the date, professional development. Here, I'll just let it finish thinking. Okay. All right. Let's see if we can agree with this. So, save the day, professional development conference for the technology industry. That's a little bland for my taste. I'm probably going to regenerate that. And they don't need to know my budget. So, this is actually a great example because this is a hallucination. It's not a great output. I expected a little bit more from the top here. But before I regenerate it, I'm going to go through and see what we can expect. Engaging keynote speakers. Yeah, I asked it to do that.

30:59Interactive panel discussions, hands-on. All right. So, this is not what I want. I'm going to say irrelevant. I'm going to submit and I'm going to try it again. Now, I want to take this to, again, the generalization of some of this stuff. You can do most of this, not all of this, on any other large language model. You can do this on Cloud2. You can do this on ChatGPT, Bard, whatever you want. On ChatGPT, a little trick that a lot of people don't know unless they've been listening to this podcast for a while. There is a very big difference in ChatGPT and in any model. whether you ask the question again, meaning you just try to correct it and run it again, versus if you regenerate the same response.

31:44And the way to do this in ChatGPT, there's a little pencil button next to the prompt that you write. And the difference is if you continue the conversation, it remembers the wrong answer that it gave you. And it's going to take that into consideration when it's doing the next thing. If you go back and use the little edit button, the little pencil thing, and change the prompt and then ask it to go ahead and generate it, it forgets the wrong answer, which you want it to forget. And so the only big difference between doing this in Jasper, other than some of the stuff that Ashley mentioned, is that here, the thumb down vote tells it in the long run that this is not what we want to get when we're trying to plan an event that ChatGPT will not remember because you're starting a brand new chat.

32:37So that's the only difference. But everything else you can really do with a very detailed prompt like this one, with a very clear process leading into this, like the one Ashley's sharing, you can do that on any other platform. Thank you. That was a good call out. So in addition to that, since I didn't like the input I thumbed down it thumbs down it that's not right I disliked it I'm going to talk that through sorry but I also don't like the next one that it generated as well it's they're getting better but just for kind of a frame of reference on Jasper you can click speed or quality I've been clicking speed this whole entire time which basically means it will just generate outputs faster but the quality might not be as good so I changed it to quality now I also clicked enhanced prompt.

33:19So this immediately forces it to give itself context so that you don't have to. Again, I didn't like the prompt two times now. I don't want to give it more context. I want it to help itself and learn. So let's see if we get a better one now. here we go okay metropolitan areas just as a reminder is the geographical location that we decided on so it gave me mccormick place which is in chicago hilton which is in canada and then another convention center in california okay great i like that it gives me like a little bit of history too once it stops generating you'll be able to see it but here we go and this is typically why I stick with the speed and not quality.

34:02I'm more of like an instant gratification type of AI say. All right. All of these days. And you'll notice this is Ashley's brand voice. Something else while we're waiting on this is you can pick which brand voice you use. So if I want it to sound like me, I actually copied in my LinkedIn profile and it analyzed my voice and tone for all of my posts. And that's my brand voice. So it's not going to sound like an AI machine. It's going to sound like me. And to pause you on that for a second, that's something I'm doing and my team is doing on ChatGPT and Claude2 as well. And the way we do this, we literally created a long prompt that tells us how we write, what we like with a few examples.

34:44We save it as a segment of a prompt in Magical that we use every time before we generate anything. So now I have Isar Personal and I have Isar business and I can use them whenever I want to use them when I don't want to sound like plain vanilla Chachipiti, but I want to sound more like Isar with my experience and my little wittiness and et cetera. And it will sound a lot more like me and a lot less like an AI. And you can do, again, you can do this here. Absolutely. It's built into the system. Jasper is amazing that way, both in means of personal branding, as well as a corporate brand that you will keep in line with, but you can do similar things with a little more work on any large language model out there.

35:28I don't know about you, but there's like certain words that just make me cringe, like synchronous and synchrony and the other like overused words. So I even take it a step further and tell it not to use those words and replace it because I wouldn't say that. I'm the same with like fancy, really formal thank yous at the end of emails and stuff. I'm like, I'm not, I'm the cheers kind of guy versus the with best amazing regards with respectfully. That's, I will never say something like that. So I tell it not to use these things and then it never does. Yep. Okay. So here are the three options and it gave me like a little bit of history.

36:05I'm personally going to choose the McCormick place and something else that you'll notice. So if you go all the way down to the bottom of this, it's got references. So that's nice. So if you don't trust it, I know sometimes people are a little bit weary, as they should be, about the fact checking. You can always ask it for references if it doesn't provide them for you. I'm going to say, and I'm just going to do this in live time so you can see what terrible prompt I put in and how it enhances it easily. So I'm going to say I like option one, plan an event, no, plan a panel discussion for this convention and provide the topics and a budget, no, and a meal plan for 100 attendees on a budget of$150 ,000.

36:59I will also mention, I misspelled so many things on here. And God bless Jasper, because it doesn't matter. It knows what I'm trying to say. And I don't even think you can expect that from people. Okay, I switched it to quality and I asked her to enhance this prompt. So again, act as an event planner, organize a panel discussion, create a meal plan. You saw the small crappy prompt I put in. So again, for the people who are just listening to this after the fact, Ashley gave it one sentence. We got two full paragraphs of a much more detailed prompt that is based on A, her understanding of who Ashley is from history, but also on this entire process that we have done together in the last 30 minutes or so.

37:41Yep. And I just asked her to create a meal plan. But when I've done this for a live event in the past, I actually put in dietary restrictions and asked it to create me meal plans based off the dietary restrictions and the budget restrictions, which is really nice. Okay. it is thinking. Hopefully not thunking. So for those of you who don't know the joke of thinking versus thunking, this comes from, oh, I'm going to butcher her name, but she is an executive at Google and her name escapes me right now, but she wrote this piece. What is it? Cassie. Yeah. Cassie. She has like a really fancy, crazy title, like the head of decision making in Google or strategic decisions in Google, which means she's a really smart cookie.

38:26And she's really an amazing person and she's really smart. And she wrote this piece about thinking versus thunking, meaning she's talking about the fact that AI is going to replace all the thunking, meaning all this stuff, right? All the tedious research and putting together documents and plans, like all the stuff that is not thinking, assisting us humans that do the thinking and replacing all the thunking that we do not want to do. And hence the mention of the thunking several times during this conversation. This is where it comes from. So thank you, Cassie, for putting terminology to what we think and not thunk.

39:06Exactly. Okay. So it created a high level agenda, which is what I asked it to do for my panelists. So an introduction and then the discussion round one, we're going to ask each panelist to share their thoughts on the current state of industry. We can either pick technology or marketing and future prospects. The panelists will dive into the innovations and technological advancements shaping the industry. Sustainable practices and responsible growth. That sounds great. And then Q &As. This sounds like your typical panel discussion and formatting. And I like that it gives me the timing breakdown.

39:36From there, I asked it to create a gourmet experience. I didn't. I asked it to create me a meal plan on a budget, but it just automatically thought and got more creative and gourmet experience sounds way better. So appetizers, an assortment of bite-sized delights, including regional ingredient-infused bruschetta and vegan spring rolls. Love it. Entrees are going to be herb-crusted salmon, stuffed ravioli, and vegan stew. Don't love the vegan stew, but that's okay. Someone else out there will. Desserts are chocolate mousse or a sorbet. And then they gave me drinks. So wine, beer, and non-alcoholic beverages.

40:10Okay. From here, all I have to do is dive into the specifics and the talk track and then come up with an invitation, send it, that says everything that this beautiful machine just created in the last 20 minutes. But I didn't, I spent 20 minutes on this. That's it. So first of all, I want to say, Ashley, this is freaking incredible. Like literally the, first of all, the thought process, it's very obvious. And I want to say something that goes back to the thinking versus thinking part of things. It comes with a lot of experience, meaning the fact you can do this now in, I don't know, 30 minutes that we have done this, is because you A, have pulled off a few events in your past, and B, has been training this particular machine to do these things again and again.

41:00So yes, there is magic, but this magic comes with knowledge and experience that you need to bring into this. Like it's garbage in, garbage out, good data in, great results out. But putting that as a given, you can do things today and now you can take this to literally any company process. It does not have to be a marketing process. This can be, how do we write better proposals to win more RFPs and go through the same process? Okay, what have we done previously that has worked? Who we need on our team? What kind of expertise we need on our team to win more business in this particular niche that we're in.

41:42Who are these people? Can we offer them jobs? Where do they live? What will make it more attractive to them? How much money they make? Literally, you can go down through a similar process with almost any business process that is in the knowledge world, meaning it's not going to put more raw material on your production floor tomorrow. It can probably help you, by the way, with supply chain management, maybe. So there is the knowledge process side of things of almost everything in business. And if you follow the same concepts that Ashley shared with us today on an incredible value on its own, but I'm saying now generalize it to any process in your business and think about, okay, we know what needs to happen.

42:28How can we build this process back and forth multiple times, but then in three weeks, four weeks, a month, a quarter, you can do this magic that used to take weeks, literally weeks involving, I don't know, four to 10 different people and cost you a hell of a lot of money because some of these people are external consultants. And now one person can do them in an hour. And let's say, okay, this was a game we were playing with a fictitious company, okay, a day. It's still comparing to weeks of multiple people is a very big difference. The other thing that I want to say, and I said that in the beginning, at the end of the day, there's people in the business and there's people in the clients, and this doesn't stand in a vacuum.

43:16Meaning if you really want to make this successful, this should be, and maybe every single step or maybe a bunch of steps together as, okay, hey, I'm going to put my team together and we're going to review two or three options that this thing gave us. And we're going to decide on one of those options and then we'll continue the process. So selecting the venue, that's a big deal because it will mean location. I mean, how would you get there? It means budget. It means travel. It means how many people can we fit in there? Either is it too big or it's like there's other things in a business that comes into putting an event together, but you're not starting with a blank piece of paper on anything.

43:53Like you're pretty far into a good direction before you have the first meeting on any of those status meetings. And to be completely fair with these tools, now these status meetings that used to take weeks can be a hackathon. Like you can literally have the relevant people in the room for one day or half a day. And at the end of that half a day, you're ready with the list of people, with the event, with what you're going to serve as food, which then, okay, now you send out to five different caterers in the Quebec area, whichever location we picked, and you get quotes and you move on from there.

44:31So it's an incredible way to take really complex, long business processes and shorten them not by 20%, but by 95 % with significantly less budget. So now I want to open it to questions. I will ask the first question and then maybe more people will jump in. how long did it take you to get from and you said years of using jasper but this particular process of finding personas and finding the people on linkedin and then creating this how long did this take before you hone it in to the i don't know if it's perfection but the very good state it is in right now um yeah i started using ai because i wasn't confident in my tone and talking and doing seconds.

45:23Like this is how I started this process was not where I'm at right now. So I just want to emphasize I did not go in and know what I was going to do with it. I went in very blind. I would say all together, it took me one time of trying it to really like understand and then build on that. I always started with just verifying in my company, we have our buyer personas. So I would ask it and test it and then I would compare it to what we have. And then as I started to trust it, I would build off of that and grow, but not long at all. So perfect. And I want to say two things on what you said. One is don't be afraid to experiment.

46:00Like a lot of people, I don't know what it's going to do. Okay, then try. Or you don't know what to do. So first of all, find maybe the most important thing when it comes to the whole AI thing is education. Find the right people. If you're listening to this podcast, it's a good step forward. I'm tapping myself on the shoulder, but find the people you trust who bring real value to you consistently, that has real experience. People like, Ashley did not do this for fun. She's really doing this thing for a business, throwing actual events for people that drive growth for her company. Find these kinds of people who are not just hype and, okay, here's a hundred amazing prompts to do this and that, but people who are real practitioners and follow them and see what they're doing.

46:47So keep on educating yourself. But there is no way of making this successful for your use case, for your business, for your personality without actually getting your hands dirty. So that's number one. Number two, when you're getting your hands dirty, do it in a safe sandbox. Ashley said in the beginning, she didn't just hit send. And okay, let's see if this even makes sense. Are these personas aligned with the real personas that we as the thinkers of the business agree that are the personas we want to go after? That we as the marketers of the business think that is the right tone to approach those people.

47:22So you have to experiment, but you have to experiment safely in the beginning, whatever that thing is. and as you develop it and develop trust in a few things, and I'll say another thing that is very important here because somebody will do this and then they're going to call me back and send me an email and say, hey, what the hell? Even when you develop trust with these platforms, they will surprise you. They will pull something stupid out of thin air and they'll make it sound extremely convincing. So it happens to me at least once or twice a week, every week, where it will sneak in something within a lot of stuff that is perfect and accurate, that is complete BS.

48:07And it will do it in the same amount of detail and the same level of quote unquote confidence as all the stuff that is real. And so you have to be aware all the time that these will hallucinate and they will pull stuff in the most unexpected places. So you still have to have the right safeguards from a business process perspective in order to make sure you're not shooting yourselves in the foot. That's why I said start safe. Start with internal projects. Okay, so I'm doing this internal project with 95 % AI. And I'm going to encourage everybody in the business involved in this internal project to use AI in this way.

48:46And if we break it up, then okay, it's an internal project. We've learned a lot. And then from there, you start with small client projects and then you continuously grow from there. There's a question from Chris on why Jasper. So I see a couple of these questions. So Jasper was actually before ChatGPT. So I don't want to just say that though, because I use Jasper over ChatGPT for many reasons. Personally, I rolled out AI in all of our marketing departments. So I was swimming in SEC compliance and data protection agreements. So just from a security standpoint, I really like Jasper. I also am a firm believer in being intentional.

49:27I love technology, but it changes every day. So I really grilled their head of product on what their product roadmap looked like and what they were going to do. Because if I'm going to encourage everyone to use AI, and I'm going to use it in my day to day, I want to know that it's not just here for the short term. I want to know that I'm going to be able to make this last and blow on debbie of it and sustainability of it was really important to me. But that was a really long way of saying why I chose to ask for over touch APT and I hope that was helpful. And I'll add on top of that, in the long run, and even now, there are two completely separate paths to all of this.

50:03One is the general large language models, which are, by the way, the underlying engine for most of the stuff you're going to use. So very few companies can develop their own models because it's insanely expensive. So that's number one. But so there's going to be that general direction and then there's going to be the optimized for X model. And that could be optimized for a specific type of medicine. And it could be specific to a specific branch of marketing. and it could be specific to writing code or maybe to writing JavaScript. There's going to be these very specialized, better trained, optimized for specific things, models.

50:47And Jasper is a great example of that, right? It's a model that is optimized for marketing, for doing the kind of stuff that Ashley's doing with this, for creating blogs and social media posts and so on. This is what it's optimized to do. So in that particular thing, it's going to perform much better than it will compared with just ChatGPT. That's number one. Number two is the user interface. As I mentioned several times during this conversation, you can do most of this with ChatGPT with a lot more work. Now, can you simplify the work with some of the tricks that I mentioned? Yes. It's still going to be more work and less user-friendly, meaning people like me, people like Ashley can probably figure it out.

51:31People are your regular, normal employee in the company, maybe in the longer run. But if you want to get quick wins, give people something that has the user interface, the ease of use, and the process that is aligned with what they're used to doing. It's going to make their life a lot easier, and your adoption success is going to be much better. Joseph, If you had something to add or to ask, please go ahead. Joseph, actually, I just want to ask you, in the beginning, do you have a framework or a formula that helps you get started? How was it like? Chaos, because it wasn't mainstream yet. No, I did not have a framework when I got started because it was not very popular.

52:12And there was not a lot of people using it. And it's a little bit more taboo than it is now. I would highly recommend using frameworks now. There's prompts everywhere. Jasper gives theirs away for free as well. They also have a free, just a basic level training. And I'm sure other AI models do as well. So if I had frameworks, I absolutely would have used them and taken advantage of them. But yeah, I did not. I will add one more thing to that. And then we're running out of time. So one thing, and I said that before, is education. Like you have to educate yourself and your team. And if it's a business, the best way to do that is to build a console, like a number of people who are in charge of AI across the business, who are consistently listening, watching, learning, and building an education plan for the rest of the people across the company on the different aspects of the company.

53:06So there's going to be different training for people in marketing than there's going to be in finance versus operations versus DevOps, right? Depending on which company you're in. But keeping the knowledge coming in and training the people consistently, because this thing is moving not just fast, but faster every single week. And so keeping on track with the latest technology and the capabilities is very important. The flip side of that, and I will say that because Ashley mentioned that, and that's critically important and very smart, which is there are probably hundreds of new AI-based tools coming out every single day.

53:47And some of them are really cool. And some of them can solve really interesting problems. At the end of the day, it's a business and there is business strategy and there is risk assessment and there is implementation across teams and processes. And that has to come first. If you just allow your employees to go wild and say, yeah, go find crazy tools and use them, that's chaos. That's real chaos. Not the chaos that Ashley was talking about. Ashley chaos. That's business chaos, which leads to really bad results. If on the flip side, you're saying, we're going to define business goals. We're going to assign these goals to a committee, to a group of people that are going to be in charge of finding AI tools.

54:28And these people will think like Ashley, meaning, is this tool, is this new startup that has this really cool tool going to be around two years from now? Because we're going to invest four months in deploying this across the company. And then two months later, they're out of business. we have a freaking problem. So in a business context, you have to think business strategy, operations, implementation, education, customer service, like all these things that come into a business before you jump in into any new cool thing, while you're evaluating every new cool thing. Because even if this company is going to fail in doing this, somebody bigger, more stable with the right roadmap, we'll take it because it's a great idea.

55:12And then maybe you want to use that tool. Yes. I want to go ahead, Ashley. No, I was just going to say, if I may leave you with one piece of advice, it would be to be intentional. I think that's a great place to land on it. And if you are in a leadership role at a company and you are trying to roll this out, trust comes from the top. So you have to be doing the same work that you're asking your employees to do. So just wanted to leave you with that. I think it's really important to mention. Ashley, one last thing before I thank you again. If people want to follow you, find you, know more about you, about your company, how do they do that?

55:49Yes. It's just Ashley A. Gross on LinkedIn. Please feel free to follow me. I love AI. I will nerd out and talk to you about it all day long. I will. I promise you I will. So thanks everybody for joining us live. So the 20 people plus we had here on the call, plus the I don't know how many people on LinkedIn Live. I really appreciate your spending the time with us. I hope you found this valuable. We're going to try to do this, and this has been successful so far. Every other week on Thursday at noon, this seems to be a good time, and a lot of people are joining every time we do this, and we're going to keep on bringing people as brilliant as Ashley.

56:26And so thank everybody for joining us from wherever we are in the world. And Ashley, thank you again for A, what you're doing, B, for sharing everything that you're doing. It's not a given. I don't take this for granted. I really appreciate you and appreciate you taking the time and sharing this with us. Thank you. Wow. Right. Incredible. Really, really incredible. Once you start thinking outside the box and you're trying to look for the places where generative AI can really do incredible things for us as far as improving processes, you're not getting a 10 % efficiency gains. You're getting 10 ,000 % efficiency gains, as Ashley just shared with us.

57:07And I really hope that you'll be able to generalize either with the comments that I provided or on your own to other areas of your business where you can do the same thing.

57:23Before we dive into this week's news, I would like to ask a small favor. If you like the podcast. If you enjoyed this episode, please share it with other people that can benefit from it and give us a review on your favorite podcasting platform that helps us reach more people and helps you be a part of a better AI world. If you found this particular episode helpful and you're looking to join us live so you can ask questions on our next live session, just connect with me on LinkedIn. We publish these as live events on LinkedIn. It happens roughly every other Thursday where we bring a practitioner like Ashley and dive into very tactical ways to leverage AI in a business.

58:01And now let's dive to this week's news. There've been some huge announcements this week, and we'll start with two that come from Microsoft. Microsoft announced two different things. One is that their Copilot 365 program is going to become available to anybody who is going to purchase it starting November 1st. So it's literally just around the corner, this is obviously huge because it means that you can have AI at your fingertips at more or less every single thing every employee does in your business. Now, are businesses ready for that? Absolutely not. I think there's very few businesses, if any, that even understand the implications of what this means.

58:40This means that writing emails, creating presentations, responding to stuff, analyzing data, all of that becomes significantly better and faster, and yet a lot riskier if there's no clear guidelines, instructions, processes, oversight, and most importantly, strategy and education for your employees. So if I have to urge one thing to you, if you are in business, in any position, either go to your management or if you are in management and say, hey, this thing is coming, we'll have it available. It has incredible benefits, but as a company, we have to put the right processes, the right guardrails, the right strategy in order to actually make this efficient versus shooting ourselves in the foot.

59:22And if you think that shooting yourself in the foot is not something that can happen, well, let's go back to Microsoft as well from this week. Two employees from Microsoft AI research team exposed 38 terabytes of private data that includes backups of their workstation as they were publishing stuff to GitHub as part of their open source effort that exposed data secrets, private keys, passwords, and 30 ,000 internal Microsoft team messages. So if you think that Microsoft does not have guardrails in place to try to prevent these kind of things, well, you're probably wrong. And now allowing AI to do some of these processes, because they're going to be at the fingertips of every single employee, think about the level of exposure it puts every single organization.

1:00:13So if Microsoft can run into a situation like this, you might too, and your organization might as well. And so putting the right strategies, guardrails, and team in place to figure this out before you allow all your employees to try using this is a great idea. On the other hand, if you do figure this out and you understand how to do this safely, you can gain a huge gain over your competition, as we've seen as an example in today's episode. Let's stay on one last piece of news on Microsoft. Microsoft just announced Windows 11, and Windows 11 is probably the most significant version of Windows maybe ever.

1:00:50And the reason is it will have AI connected to everything. So just like they announced co-pilot for Microsoft Office, they're now announced co-pilot for everything Windows, meaning you'll be able to use the power of generative AI with everything within your computer. You'll be able to ask questions about your data. You'll be able to produce documents. You'll be able to produce imagery. You'll be able to remove backgrounds from screenshots. You'll be able to summarize documents, web pages, conversations, anything you can imagine because it's going to be baked into the actual operating system of your computer.

1:01:24So take what we said on Microsoft 365 and now quadruple that because it's going to be connected to literally everything on your computer. And that's coming out now on Windows 11. So buckle up. It's going to be a very, very interesting ride. Another big player that made a huge large language model announcement is Amazon. Amazon just announced that Alexa is going to be powered by a new large language model that Amazon claims is going to provide its superpowers that even ChatGPT doesn't have because in addition to just be able to answer questions, you'll be able to do things for you because of the things Alexa is already connected to because it has access to the internet and so on.

1:02:04This is not surprising. It's obviously interesting news. And I've mentioned that several times on the show before. I expect Siri and Alexa and Google Home to do all these things just the same and take those large language models, combine them to everything they're doing today, combining it with all the devices they're connected to, and allow you to be really conversational with everything you need in your house. And with these personal assistants, I think in the very near future, within months, we'll be able to do really magical things with them. And the first one that made the announcement that it's becoming available is Amazon.

1:02:35Another huge announcement this week is OpenAI just unveiled DAL-E3. DAL-E is their text-to-image generator. It actually came out before ChatGPT in 2022. DAL-E3 is coming out imminently. It's rolling out to the paid members as we speak. And the biggest difference between that and all the previous models is that it's going to be a part of ChatGPT, meaning it's really a multi-modal capability that allows you to chat back and forth while you're creating the image. This is a huge benefit compared to using existing system like Stable Diffusion or DALI 2 or even MidJourney. And there's two main reasons.

1:03:11Reason number one is that you can write much longer and more detailed prompt compared to what you can do with a standalone text to image generators. But the other reason is a lot deeper and a lot more profound is the fact that it understands the context and it's a part of the conversation, meaning it understands roles, it understands what you're trying to create, It understands the part of the project that it's in, and you can go back and forth and adding and changing and manipulating the image because it's a part of a conversation. In addition, they obviously improved the model, so now you can create much better images than you can create with DALI 2 before.

1:03:46So I expect DALI 3 to be a huge hit. I'm personally really excited to start playing with it as soon as it comes out and compare it to what currently is my number one pick mid-journey. I will keep you posted as I test things out. And the last really big piece of news that I'm going to share with you today is that rumors are hinting that Google is getting ready to release Gemini. Gemini is their nuclear version of large language model battle with OpenAI. That's what's supposed to put them back on the map. Everything Google has done so far in the AI battle against OpenAI has been, in simple terms, disappointing.

1:04:22I mean, BARD's capabilities do not come even close to what ChatGPT or even Claude2 from Anthropic. and even on the web side of things, I find that Bing is actually doing a better job across multiple things. So very big disappointment from Google so far. That being said, Google has access to more data than anyone and definitely budgets for both the best people and the best chips on the planet. And hence, I am certain that what they're baking is very, very interesting. Gemini is said to be a real multimodal AI model that will be able to handle graphs and images and text, et cetera, in a very good way.

1:05:00And it also leverages, as we mentioned, Google's access to a lot more data than anybody else has, including stuff that nobody else has access to, such as data from Android devices and YouTube. So I expect Gemini to finally put Google at par with OpenAI. And the reason I want that is I want us all to have options of different tools to choose from because it will give us end users more options to choose from. And over time, we'll probably make it more accessible and cheaper for us to use advanced capabilities. That being said, I'm terrified with how fast these things are moving. And the fact that this is another quantum leap forward by another really big player will force everybody else to move forward even faster without any real guardrails at this point.

1:05:45So with this not so positive note, but the reality of where we are in the AI world today, I encourage you to go and test AI systems, try different use cases, learn about it as much as you can, share it with other people, because if we keep on sharing, we help the world, we help society get more ready for this tsunami that is coming. And until next time, have an amazing week.